Case study
Down the Hall, Never Booked: The Internal Referral Nobody in the Clinic Owns
A multidisciplinary clinic's cheapest appointment is the one a practitioner refers to a colleague thirty feet away, and it is the one most likely to evaporate. This is the transformation that catches the referral where it is made, carries only what the patient agreed to let travel, and escalates the ones that go quiet.
No client engagement behind this piece: it is how this category of product would be transformed, with benchmark-sourced targets.
Who this is for
A typical independent multidisciplinary clinic
Signed off by Clinic owner or managing partner, with the practice manager scoping
Industry
Private healthcare (multidisciplinary and integrated care), 1 location, 14 practitioners across five disciplines, about 9 administrative and support staff
Legacy stack
Engagement
Concept
Contents
The situation
A patient sees the physician for low back pain that has not settled in six weeks. The physician examines them, rules out what needs ruling out, and says the sentence that is the whole reason the clinic exists in this shape: “I want you to see our physiotherapist.”
Thirty feet away, there is a physiotherapist with capacity on Thursday.
What happens next is almost always one of three things. The patient stops at the front desk and books, which is the good case and happens less often than anyone believes. The patient says they will call, and does not. Or the patient books, arrives on Thursday, and spends the first eleven minutes of a forty-minute appointment telling the physiotherapist the same history they told the physician, because nothing travelled with them except a name in a calendar.
A multidisciplinary clinic is built on the assumption that the second appointment happens. It is the reason the disciplines share a lease. And the handoff that produces it is, in most clinics, a sentence spoken in a treatment room and remembered by nobody.
Awaiting booking. Front desk actions it, nothing books itself.
What travels
Not booked by day 5, it goes to a named person. One referral, one owner.
"I want you to see our physiotherapist."
No record of any of them. The ones that never book are not late, they are invisible.
What the evidence says, and what it does not
There is no published figure for how many internal referrals inside a Canadian multidisciplinary clinic reach a booked appointment. Nobody measures it, which is itself the finding: a number nobody tracks is a number nobody manages.
The closest rigorous evidence is about referral loops between organisations. Patel and colleagues, publishing in the Journal of General Internal Medicine in 2018, analysed 103,737 referral scheduling attempts in a large health system over one year. 36,072 of them, 34.8%, resulted in a documented completed appointment.
That number is not this clinic’s number and this study does not pretend it is. It is American, it is one large health system, and it describes referrals sent between organisations, where distance, insurance and unfamiliarity all work against completion. Every one of those frictions is absent when the physiotherapist is thirty feet away.
Which is the point worth taking from it. The referral loop leaks badly even when a health system has an EMR, a referral module and staff whose job is to close it. A clinic whose internal referral mechanism is a spoken sentence should not assume it is doing better because the walk is shorter.
The design
The practice management system stays the source of truth, and nothing is migrated.
The referral is captured where it is made. Not at the front desk afterwards, not in a message the practitioner writes at the end of the day. At the moment the physician says it, a referral exists with the receiving discipline, the reason, and the clinical context that makes the first appointment useful. That is the single highest-value change, and it is mostly an interface problem rather than an AI one.
What travels is scoped, per practitioner. The receiving practitioner does not open the chart. They open a summary that answers what the next clinician needs to know: presenting complaint, relevant findings, what has been ruled out, what has been tried. What may appear in that summary is governed by what the patient has agreed can travel, recorded per practitioner rather than per role, because that is how PHIPA actually works in a building where each practitioner is their own custodian. An express instruction to withhold is enforced by the system, not remembered by a receptionist.
Nothing books itself. The referral becomes a request that a person actions. Front desk staff know which practitioner a given patient will actually return for, and which Thursday slot is really available. The software’s job is to make sure the request is visible, not to make the decision.
The ones that go quiet escalate. A referral not booked within the clinic’s own window, five days by default, is surfaced to a named person rather than aging invisibly. This is the part that recovers revenue, and it works only because of the first change: you cannot chase a referral that was never written down.
Where the return lands
Two places, and a clinic measures them differently.
The first is the retained referral. An internal referral that converts is the cheapest appointment the clinic will ever book: no marketing, no acquisition, no wait, and a patient who already trusts the building. Break-even for the first slice is around twelve retained referrals a month at the clinic’s own average revenue for an initial visit in the receiving discipline, which the clinic supplies rather than takes from here.
The second is the eleven minutes. A patient who confirms a history rather than reconstructing it gives the receiving practitioner most of a first appointment back. That time is clinical, not administrative, which is why it is worth more than the same eleven minutes saved at the front desk.
What we would refuse
- We will not auto-book across disciplines. A booking made without a person is a booking the patient did not choose, and in a clinic that bills across four different rails it is also a billing problem waiting to happen.
- We will not build a single merged chart. It is the thing clinics ask for first and it is the thing PHIPA is least comfortable with when the practitioners are separate custodians, particularly when some of them are contractors. Scoped sharing is more work to design and is the honest structure.
- We will not let the summary paraphrase clinical findings. It carries what was recorded, attributed, with the gaps visible. A model that smooths a missing measurement into readable prose has invented a finding, and the receiving practitioner has no way to know which sentence to distrust.
- We will not start with all five disciplines. One referring discipline, one receiving discipline, one metric. The pair with the highest volume is usually the pair with the worst leak, and it is the one that pays for the rest.
Privacy
Because records stay in the clinic’s own environment and processing runs with zero data retention, the design lines up with what PHIPA expects: a privacy impact assessment before deployment, an agreement naming the processor as an agent under section 17, and an audit log on every access to personal health information.
This segment carries one question the others do not. When several practitioners in one clinic are each their own custodian, a system whose purpose is to let them share electronically raises the question of whether the vendor is a health information network provider under Ontario Regulation 329/04, which brings duties beyond an agent agreement, including a threat risk assessment and an access log made available to the custodians. That question is settled at scoping. A clinic should take its own legal advice rather than take this from a case study.
The outcomes
Internal referral from spoken to booked or escalated
Verbal, tracked nowhere Surfaced to a person at day 5
Untracked to 5 days1
Design targetReferrals reaching a documented completed appointment
34.8%2
Industry benchmarkBreak-even, in retained internal referrals per month
About 123
Design targetHistory a patient repeats at the second door
Retold in full Confirmed, inside consent
Retold to confirmed4
Design targetTime to the first shippable slice
4-6 wk5
Design target1 Design target measured against the recreated legacy handoff shown in the before/after above. Five days is the clinic's own setting, not a fixed rule: it is the point at which a referral that has not been booked stops being in progress and starts being lost.
2 Patel et al., 'Closing the Referral Loop: An Analysis of Primary Care Referrals to Specialists in a Large Health System', Journal of General Internal Medicine 33(5):715-21, 2018: of 103,737 referral scheduling attempts between July 2015 and June 2016, 36,072 resulted in a documented completed appointment. US, one large health system, and primary-to-specialist referrals BETWEEN organisations rather than between practitioners inside one clinic. It is quoted to size how badly referral loops leak when nobody owns them, not as a Canadian multidisciplinary number, and this study does not invent one.
3 Design-target arithmetic, not a measured result. The first slice is a fixed fee; break-even is stated as the number of internal referrals per month that would otherwise have evaporated, valued at the clinic's own average revenue per initial visit for the receiving discipline, which the clinic supplies.
4 Design target for the handoff described in the how-it-works section. What crosses is scoped per practitioner rather than by role, so the receiving practitioner opens a summary the patient has agreed may travel and confirms it rather than rebuilding it.
5 Standard first-slice scope: one referring discipline, one receiving discipline, one metric, evals and a human approval gate included.
Frequently asked questions
Does this book the appointment automatically?
No, and that is deliberate rather than cautious. The referral is captured the moment the practitioner makes it, with the reason and the relevant history attached, and it becomes a booking request rather than a booking. Front desk staff know things the software does not, including which practitioner this particular patient will actually come back for. What changes is that the request exists somewhere other than a practitioner's memory, and that one nobody has actioned in five days is surfaced to a person instead of expiring quietly.
Can the receiving practitioner see the referring practitioner's whole chart?
No, and this is the part that is design work rather than a settings screen. Each regulated practitioner in the clinic is their own health information custodian under PHIPA, so a shared record is several custodians sharing rather than one filing cabinet. Sharing for the purpose of providing care is permitted inside the circle of care on assumed implied consent, and a patient may expressly withhold it from any one practitioner. What crosses on a referral is a scoped summary rather than the chart, the scope is held per practitioner rather than per role, and a withheld instruction is enforced by the system rather than remembered by staff.
Our practitioners are contractors, not employees. Does that change anything?
It changes the privacy structure rather than the workflow, and it is the most common shape in this segment. A contractor practitioner is typically their own custodian rather than an agent of the clinic, which means the clinic is not automatically entitled to their records and cannot simply grant access across the group. The agreements have to say who holds what before the software moves anything, so that is scoped at the start of an engagement rather than discovered during it.
Is a shared record across practitioners still PHIPA compliant?
It can be, and the honest answer is that it depends on structure rather than on software. The design keeps the practice management system as the source of truth, records what the patient agreed to per practitioner, logs every access, and completes a privacy impact assessment before anything touches a live chart. Where a clinic connects several custodians primarily so they can share electronically, there is a further question about whether the vendor becomes a health information network provider, which carries its own duties. That question is asked at scoping, not at delivery, and a clinic should confirm its own position with its own counsel.
What would this cost to try?
A fixed-fee first slice on the clinic's highest-volume internal referral pair, which is usually physician to physiotherapy or physiotherapy to massage therapy. It ships in four to six weeks with its own metric, and break-even is stated in retained referrals per month rather than in licences.
Go deeper on the method
Where to Start With AI in a Clinic: How to Pick Your First Workflow
The hardest part of AI in a practice is not the model, it is choosing where to begin. A simple way to pick a first workflow that is high-value, low-risk, and proves itself fast, and the three candidates a clinic should never start with.
Read article
Your Patients' Data Stays Yours: Adding AI Without Handing Over the Chart
The top reason a practice stalls on AI is not cost, it is not knowing where the chart goes. Here is how to add intelligence while personal health information stays in your environment and in Canada, plus the seven questions to ask any vendor before you sign.
Read articleMore transformations
The Refill Line: Turning the Pharmacy's Loudest Conversation Into Something You Can Work in Batches
A community pharmacy answers the same question forty times a day, one interruption at a time, in the middle of dispensing. This is the transformation that lets the request arrive as structured intake instead of a phone call, keeps a pharmacist signing every fill, and measures the change at the counter rather than in a dashboard.
21.7%
Ontario pharmacy professionals naming workload as their top driver of burnout
Stop Resubmitting, Start Fixing: Reading the Rejection Pattern Instead of Working the Queue
A billing operation measures itself on how fast it reworks rejected claims, which rewards being efficient at a problem instead of ending it. This is the transformation that groups rejections by cause, sends the fix upstream to the clinic, and keeps a human on every resubmission.
$83B
Annual staff time spent on routine administrative transactions
Get the next transformation in your inbox.
When there is something worth your time, you will be first to know. No spam, unsubscribe anytime.